Close Window

University of California at Santa Barbara: "Investigation of mesoscale wind patterns in the California bight: Influence on extreme precipitation events"

Final Report



Coastal Southern California provides unique challenges to modelers and forecasters, alike. To the south and west are the cool waters of the Pacific Ocean. To the north and east-beyond the mountains-lies the Mojave Desert. On many days, the boundary between the cool, moist marine air from the Pacific and the hot, dry air of the deserts lies somewhere in the inhabited coastal valleys and basins where gradients of both temperature and humidity can be extreme. In addition, from a complexity standpoint, the terrain of Southern California is second to none. Los Angeles County alone encompasses the islands of Santa Catalina and San Clemente-20 to 50 miles offshore in the Pacific Ocean, the broad expanse of the L.A. Basin and San Fernando Valley, the Santa Monica Mountains that reach over 3000 feet, the San Gabriel Mountains that exceed 10,000 feet, and the Antelope Valley of the Mojave Desert. The objectives of the project were:

a) Establish a local PSU/NCAR MM5-V3 capability at the WFO.
b) Investigation of recent flood events occurred in Southern California during the El Nino of 1997-1998.

It is very interesting to look back at what was planned versus what was actually accomplished. In some cases, we were able to accomplish a lot more than we had anticipated. In other areas, frankly, not as much was accomplished as we'd hoped. Overall, we learned a great deal about mesoscale modeling while developing a useful tool that was actually used operationally during the recent rainy season. In fact, MM5 output directly contributed to the timely issuance of a flash flood warning for portions of Ventura and Los Angeles County in January 2001.

The first phase of the project was to establish a local MM5 capability at the WFO. The goal was to give forecasters a local model that could be used to forecast wind and rainfall patterns as influenced by complex terrain. This phase took considerably longer to accomplish than originally planned. Due to a death in the family, the student partner that was to accomplish the initial setup of the MM5 at the WFO had to leave school. As a result, much of the programming was accomplished by a WFO programmer, Chris Jacobson, who knew nothing about MM5 at the beginning of the project. It took until June of 2000 for us to get Chris into the MM5 Tutorial Course at Boulder, Colorado. Once schooled, it took from June to the beginning of the winter rainy season to get MM5 running reliably on the WFO workstation, successfully display the data, and get the output ported to AWIPS. With the cooperation of Dr Jones, we eventually succeeded. Today, our LOX forecasters now have two MM5 runs per day for use in forecasting both high winds and QPF with the output available on AWIPS. These runs were very useful in forecasting the heavy rains that hit Southern California earlier this year.

Besides the local WFO MM5, our partner has been running MM5 daily at UCSB since early 2000. Dr. Jones coordinated much of the setup of the UCSB MM5 to support WFO operations. For example, the inner domain is centered directly over the local LOX County Warning Area (CWA). The UCSB MM5 runs once per day providing forecasts of surface winds, temperatures, relative and specific humidity-as well as simulated radar images-at one-hour intervals out to 48 hours. These data are available to the LOX forecasters via a website, again designed by Dr. Jones in coordination with the WFO.

The LOX forecast staff has been using the UCSB MM5 output for over a year now. For the most part, forecasters have been primarily interested in the MM5 surface wind forecasts for assistance in forecasting strong Sundowner winds along the south facing slopes of the Santa Ynez Mountains, the humidity fields for forecasting the marine layer and the Catalina Eddy, and the simulated radar images for qualitative precipitation forecasting. The high resolution of the UCSB MM5 data has proven far superior to NCEP models in forecasting these phenomena.

Of particular interest are the 1-hour resolution, simulated radar forecasts. Basically, the UCSB MM5 simulated radar output gives LOX forecasters an hour-by-hour picture of what the radar should be looking like almost two days in advance. Although it is very difficult to evaluate this objectively, some of the resulting forecasts have been phenomenally accurate-including one forecast of a highly unusual early morning [4 AM] isolated thunderstorm that dropped significant hail over southern Los Angeles County and northern Orange County. MM5 hit this isolated thunderstorm event over 30 hours in advance!

The second phase of the project was to go back and study archived events from the El Nino rains of '97-'98. Due to the delay encountered in getting the WFO MM5 running, it has not been possible to accomplish this research at the WFO. In fact, the WFO MM5 has been running "operationally" since the beginning and will likely not be taken offline for research until the summer dry season. However, Dr. Jones has been able to accomplish a study of one of the heaviest rain events to hit Southern California during the recent El Nino. See section 3 for details.


Probably the biggest lesson learned had to do with being extremely flexible. Things came up in the project that we had no way of knowing in advance-like a death in the family; things we knew about had a greater impact than we supposed-like the installation of AWIPS; things that we simply could not plan for-like the loss of access to radar data archives due to simultaneous failures of redundant recorders; and, finally, the difficulty of trying to be in two places at once-which is what seems necessary when everyone working the project is so very busy. Due to the flexibility, resourcefulness and dedication of the people involved in the project, most of these difficulties were overcome. See Section 4 for details regarding scientific lessons learned.


Project planning was done in a series of joint meetings held either at the University in Santa Barbara or at the WFO in Oxnard. During one of these meetings, Dr. Jones was able to meet with Andy Edman, WR/SSD. However, most of the coordination was done via email between Dr. Jones, David Danielson, Brent Bower [WFO Hydrologist], and David Gomberg [Senior Forecaster]. In addition to these, the following exchanges were accomplished:

  1. Both Dr. Jones and Brent Bower both attended the National AMS Conference in Albuquerque, New Mexico, where Dr. Jones presented his poster session on the project.

  2. Dr. Jones attended a seminar held at the WFO with a presentation by Dr. Clive Dorman of Scripps. Dr. Dorman made a presentation and demonstration on super critical flows in the California Bight and their influence on the marine layer.

  3. Chris Jacobson, journeyman forecaster at LOX and drafted MM5 expert, put together a COREL Presentations seminar for all WFO forecasters on the LOX MM5 modeling effort-including displays that are available on AWIPS. This presentation was incorporated into the local forecaster training program.


In October 2000, Dr. Jones submitted a paper based on his work with the MM5 for a flooding event that occurred in February 1998. This paper was titled "Forecast Skill of the Penn State/NCAR MM5 Mesoscale Model during the Heavy Precipitation Event of 23-24 February 1998 in Southern California." This paper contained a statistical analysis of the MM5 model QPF versus the actual rainfall for the event. The WFO provided both background information-including several Western Region research papers written by WFO personnel-and rainfall data used for comparisons with the model output. The paper was submitted to Weather and Forecasting, but was rejected due to the lack of a qualitative evaluation of model skill. It is the intention to revise and resubmit this manuscript.


4.1 UCSB
On the UCSB end of the project, Dr. Jones has benefited from the interaction with the NWS forecasters on two areas. First, a lot of information about local circulations responsible for extreme rainfall, flash floods and intense winds (Sundowners and Santa Ana's) were provided by the forecasters. The daily experience of the forecasters and knowledge of previous events have definitely provided an extremely helpful guidance for Dr. Jones to direct the research component of the project. Second, the NWS forecasters have been providing guidance in the selection of the meteorological parameters that are most frequently used in their daily analysis and forecasters. This information is being incorporated in the UCSB MM5 high-resolution web site with the goal of providing additional sources of forecast information that can be daily used by the NWS forecasters.
The primary problem encountered on the UCSB side was the lack of time and the need of a computer student assistant to help with the development of the UCSB web site.

4.2 NWS
4.2.1 The benefits from the NWS Forecaster's standpoint are as follows:

a. One of the most important benefits of this project was that it gave us the opportunity to work with a local university researcher on a project of our choice that had direct impact towards improving forecast operations. The complex terrain of Southern California produces wind currents and gradients of temperature and relative humidity that cannot be resolved by other models currently available via AWIPS. Now we have an in-house mesoscale model, displayable on AWIPS, that shows skill in dealing with these complexities. Further, this model has already proven itself in several critical weather situations.

b. We have learned a lot about mesoscale modeling. We have already had to adjust several of the parameters of our MM5-such as convective parameterization-in order for it to produce worthwhile output. Each time we do this, we learn more about how models work.

c. LOX forecasters had the benefit of the UCSB MM5 model via the UCSB website since the middle of the rainy season of '99-'00. This resulted in improved forecasts of wind and rain events. Specifically, model output from the UCSB MM5 was used to support the issuance of a flood warning for eastern Ventura County and Western Los Angeles County during the heavy rain event of January 2001.

d. LOX forecasters had the benefit of the LOX MM5 via AWIPS for much of this winter rainy season. Output from this model was used in early March 2001 to issue another flood warning for the Santa Monica Mountains area that also verified.

e. This project laid the groundwork for the next proposal-which will be to model rainfall run-off in complex terrain. This is an essential step for improving flash flood forecasting.

f. Finally, the project benefited from the sponsorship of COMET. For years this office has wanted to develop an internal mesoscale modeling capability. This could not have been accomplished without the sponsorship of COMET or the cooperation of Dr. Jones at UCSB.

4.2.2 Problems Encountered

a. The loss of the university student assistant made it necessary to develop some of the MM5 expertise in-house. Due to the lack of quotas at the MM5 school in Boulder, it took most of a year to get someone up to speed. However, we had an outstanding individual come in and develop the skills necessary to make the project work. Many thanks go the Chris Jacobson, journeyman forecaster extraordinaire.

b. We totally underestimated the impact the installation of AWIPS would have on this project. This was a much higher priority project that drained time and resources from this partnership project.

c. We had lousy timing and luck with the radar archives. A quiet couple of weeks between the southwest monsoon and the fall rainy season was set aside to do much of the radar investigation. However, when it came time to examine the archives, both archive devices had failed and could not be replaced for several weeks.

d. Early in FY2001 we realized that we need additional software to be able to display the MM5 output. Unfortunately, this came at a time when we had no authority to spend. This delayed our ingesting of the local MM5 into AWIPS by 2 months.

e. We were unable to be in two places at the same time.